DataViz Tutorial in Python using Plotly¶

In [1]:
import pandas as pd
import plotly.express as px

Check Plotly's version.

In [2]:
import plotly
print(f'plotly version={plotly.__version__}')

plotly version=5.18.0

Read and preprocess data¶

In [3]:
df = pd.read_csv('https://github.com/subwaymatch/mba562-dataviz-demos/raw/main/dataviz-tutorials/superstore_data.csv')
df
Out[3]:
Row.ID Order.Date Sales Category Sub.Category Region
0 1 11/8/16 261.9600 Furniture Bookcases South
1 2 11/8/16 731.9400 Furniture Chairs South
2 3 6/12/16 14.6200 Office Supplies Labels West
3 4 10/11/15 957.5775 Furniture Tables South
4 5 10/11/15 22.3680 Office Supplies Storage South
... ... ... ... ... ... ...
9989 9990 1/21/14 25.2480 Furniture Furnishings South
9990 9991 2/26/17 91.9600 Furniture Furnishings West
9991 9992 2/26/17 258.5760 Technology Phones West
9992 9993 2/26/17 29.6000 Office Supplies Paper West
9993 9994 5/4/17 243.1600 Office Supplies Appliances West

9994 rows × 6 columns

Convert the "Order.Date" column into a Datetime type.

In [4]:
df['Order.Date'] = pd.to_datetime(
    df['Order.Date'],
    format='%m/%d/%y'
)

df.head()
Out[4]:
Row.ID Order.Date Sales Category Sub.Category Region
0 1 2016-11-08 261.9600 Furniture Bookcases South
1 2 2016-11-08 731.9400 Furniture Chairs South
2 3 2016-06-12 14.6200 Office Supplies Labels West
3 4 2015-10-11 957.5775 Furniture Tables South
4 5 2015-10-11 22.3680 Office Supplies Storage South

Check the data types of each column.

In [5]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 9994 entries, 0 to 9993
Data columns (total 6 columns):
 #   Column        Non-Null Count  Dtype         
---  ------        --------------  -----         
 0   Row.ID        9994 non-null   int64         
 1   Order.Date    9994 non-null   datetime64[ns]
 2   Sales         9994 non-null   float64       
 3   Category      9994 non-null   object        
 4   Sub.Category  9994 non-null   object        
 5   Region        9994 non-null   object        
dtypes: datetime64[ns](1), float64(1), int64(1), object(3)
memory usage: 468.6+ KB

1. Line Chart: Sales Over Time¶

We’ll begin with a line chart that shows how sales have trended over time. This is useful to identify seasonality or trends.

First, create an aggregated DataFrame that shows the total sales by month.

In [6]:
df['YearMonth'] = df['Order.Date'].dt.strftime('%Y-%m')
In [7]:
df_sales_by_month = df.groupby(by='YearMonth', as_index=False) \
    .agg({'Sales': 'sum'})

df_sales_by_month.head()
Out[7]:
YearMonth Sales
0 2014-01 14236.895
1 2014-02 4519.892
2 2014-03 55691.009
3 2014-04 28295.345
4 2014-05 23648.287
In [8]:
fig = px.line(
    df_sales_by_month,
    x='YearMonth',
    y='Sales',
    template='simple_white'
)
fig.show()

2. Line Chart: Sales Over Time by Category¶

The sales peak end of 2017. We can explore whether this is driven by a specific category or region. Imagine you are a category manager and want to plot the sales for various cateories instead.

First, let's plot line chart again and then see if there are better ways to show category sales:

In [9]:
df_sales_by_month_category = df.groupby(
    by=['YearMonth', 'Category'],
    as_index=False
).agg({'Sales': 'sum'})

df_sales_by_month_category.head()
Out[9]:
YearMonth Category Sales
0 2014-01 Furniture 6242.525
1 2014-01 Office Supplies 4851.080
2 2014-01 Technology 3143.290
3 2014-02 Furniture 1839.658
4 2014-02 Office Supplies 1071.724
In [10]:
fig = px.line(
    df_sales_by_month_category,
    x='YearMonth',
    y='Sales',
    color='Category',
    template='simple_white'
)
fig.show()

Alternative ways to show category¶

In [11]:
df_sales_by_category = df.groupby(
    by=['Category'],
    as_index=False
).agg({'Sales': 'sum'})

# sort by sales in descending order
df_sales_by_category.sort_values(
    by='Sales',
    ascending=False,
    inplace=True
)

df_sales_by_category.head()
Out[11]:
Category Sales
2 Technology 836154.0330
0 Furniture 741999.7953
1 Office Supplies 719047.0320
In [12]:
fig = px.bar(
    df_sales_by_category,
    x='Category',
    y='Sales',
    color='Category',
    template='simple_white'
)
fig.show()
In [13]:
fig = px.pie(
    df_sales_by_category,
    names='Category',
    values='Sales',
    color='Category',
    template='simple_white'
)
fig.update_traces(textposition='inside', textinfo='percent+label')
fig.show()

Another type of chart that displays relativity is a treemap.

In [14]:
fig = px.treemap(
    df_sales_by_category,
    path=['Category'],
    names='Category',
    values='Sales',
    template='simple_white',
)

fig.show()

Add custom labels.

In [15]:
labels = [f"{row['Category']}<br>{'${:,.0f}'.format(row['Sales'])}" for _, row in df_sales_by_category.iterrows()]
In [16]:
fig = px.treemap(
    df_sales_by_category,
    path=['Category'],
    names='Category',
    values='Sales',
    template='simple_white',
)

fig.update_traces(labels=labels)

fig.show()